Web Analytics: Are They Really Accurate?
If you checked your site analytics, you could probably give me an exact number. But that number may not be anywhere near accurate.
Getting accurate data is surprisingly difficult, even for a company as well-equipped as Google. If you run multiple analytics tools on your site, you’ll notice huge variations in the metrics you get back.
Once you understand why this happens, you’ll have a better idea of how to interpret trends in data rather than focusing on numbers alone.
Why Aren’t Analytics Exact?
Every time a page is visited, a file is downloaded or an image is viewed, the server tracks that activity in a basic log file. But log files are really only used when a problem arises. They aren’t designed to help you understand visitor behavior.
Analytics software is more advanced, and there are two basic types.
- Log File Parsers such as AWStats process the log file and turn it into human-readable data. These scripts normally run on a server, but some are downloadable applications or can sit on an intermediary, such as CloudFlare.
Neither system is perfect.
Most people would define a ‘visitor’ as a person who visits their site on a given day. But multiple people can use the same IP address, and the same person can have multiple IPs.
Analytics software has to try to convert a mish-mash of data into a meaningful ‘visitor’ – a single human being.
Consider these problems:
- Should our analytics tool assume that multiple views from the same IP is a single visitor? If so, how long should it wait before resetting the counter for that visitor?
- Are search engine bots visitors? If not, how do you tell the difference between bots and people?
The interpretation of terminology can massively affect the data you see. For example, the bounce rate (the percent of visitors who leave after just one page view) will be drastically different on different pages, depending on how your analytics program defines a visitor (or a bounce).
Overcoming Analytics Problems
There’s no such thing as a clear answer, but if you change the way you use analytics, you’ll be less dependent on the numbers.
Instead of counting visits, see analytics software as a reporter of trends.
Even though you might not know how many people visited on Tuesday, you will probably know if there were more people on your site the day before. Comparing like with like, you start to see meaningful patterns that you can act on to improve your site.
Join our newsletter & be first to hear when we publish new posts.